Quality Control, Robust Design and the Taguchi Method

In 1980, AT&T Bell Laboratories researchers conducted an experiment with an orthogonal array design of size 18 (OA18) for optimization of an LSI fabrication process. As a measure of quality, the quantity "signal-to-noise" ratio was to be optimized. Since then, this experimental approach has been named "robust design" and has attracted the attention of both engineers and statisticians. My colleagues at Bell Laboratories have written several expository articles and a few theoretical papers on robust design from the viewpoint of statistics. Because so many people have asked for copies of these documents, it has been decided to publish them in book form.